Implementation of Coupled Skin Temperature Implementation Analysis and of Bias Coupled Correction Skin Temperature in a Global Analysis Atmospheric and Bias Data Correction Assimilation in a System Global Atmospheric Data Assimilation System Jon D. Radakovich, Michael G. Bosilovich, Michael Jiun-dar G. Chern Bosilovich, and Arlindo Jon D. da Radakovich, Silva Jiun-dar Chern, Arlindo da Silva and Francis Verter Global Modeling and Assimilation Office Global NASA Modeling Goddard and Space Assimilation Flight Center Office NASA Greenbelt, Goddard Space MD 20771 Flight Center
Motivation: Introduction surface skin temperature (Ts) is a critical state because it reflects the surface radiative properties and energy budget and can dictate convective initiation. Reliable Ts field from the operational GMAO DAS (Global Modeling and Assimilation Office Data Assimilation System) is a key requirement from scientific instrument team users. Method: integration of the state-of-the-art NCAR Community Land Model (CLM) version 2.0 land-surface model (Dai et al. 2002; Zeng et al. 2002; Bonan et al. 2002) into GEOS4. Ts analysis and coupled bias correction
Models & Assimilation System The Finite-Volume GCM (fvgcm) Capable of resolving atmospheric motions from meso- to planetary-scale with a terrain-following Lagrangian controlvolume vertical coordinate system (Lin and Rood 1999). NCAR CAM2 physical parameterizations The Goddard Earth Observing System (GEOS4) The fvdas first utilizes a Statistical Quality Control (SQC) system to screen the observational data. The analysis is then performed by the Physical-Space Statistical Analysis System (PSAS; Cohn et al. 1998).
Models & Assimilation System The Community Land Model (CLM v2) One-dimensional point model that uses sub-grid scale tiles. One vegetation layer with a photosynthesis-conductance model to realistically depict evapotranspiration (Bonan 1996). 10-uneven vertical soil layers with water, ice, and temperature states in each layer. Up to five snow layers depending on the snow depth with water flow, refreezing, compaction and aging allowed. Utilizes two-stream canopy radiative transfer, the Bonan lake model (1996), topographic enhanced streamflow based on TOPMODEL (Beven and Kirkby 1979), and turbulence is considered above, within, and below the canopy.
Skin Temperature Assimilation and Coupled Bias Correction Incremental Bias Correction (Daily): A variant of the Dee and da Silva (1998) BC scheme was implemented where, a running estimate of the bias is calculated based on Ts observations and added to the energy balance at every timestep to counteract the subsequent forcing of the analyzed Ts back to the initial state. δw w b k a f a = = w = K ( w b f f k 1 o b Hw f k 1 γ δw f + δw a + a b f ) f = b f b τ f b was added to the energy balance for the canopy and for the top layer soil/snow surface, and then updated at the analysis times.
Skin Temperature Assimilation and Coupled Bias Correction Diurnal Bias Correction (Diurnal, DBC) IBC removed the time mean bias, but not the bias in the mean diurnal cycle thus DBC was implemented. Time-dependent bias modeled with a sine wave parameterization: b f ( t) = ( a cosω t + b sinω ) a = a γδw a cosω t b = b γδw a sinω t 2π where ω 1 = is the diurnal harmonic 24h
Summary of Experiments Coupled Ts technique was tested in the fvdas CLM2 framework and run at 1 x 1.25 horizontal resolution with 55 vertical levels from May through September 2001. The atmospheric analysis is performed every 6 hours, while the Ts analysis is done every 3 hours. The bias estimate is updated at the analysis time. ISCCP (International Satellite Cloud Climatology Proect; Rossow and Schiffer, 1991) 3-hourly Ts observations were used for the surface analysis and BC.
Summary of Experiments Control (CTL) July-September 2001 Daily Bias Correction (EXP1) Includes a single bias at each grid point Diurnal Bias Correction (EXP2) Includes 3 hourly (8) biases at each grid point ANA Analysis cycle BKG Background (model) CTL-AI, EXP1-AI, EXP2-AI add the analysis increment into the surface energy budget
Analysis and Bias Correction 320 Analysis and Bias Correction 315 310 Ts(K) 305 300 295 290 0 3 6 9 12 15 18 21 24 Hour CTL CTL-AI EXP2 EXP2-AI ISCCP The analysis increments immediate impact, not long lasting, bias correction is applied at every time step
Analysis Skin Temperature Differences (July 2001Day and Night) Analysis compared back to ISCCP Daily bias corrections remove of many of the regional biases Diurnal bias corrections address most bias
Background Skin Temperature Differences (July 2001 Day and Night) Most improvement in skin temperature occurs with daily bias correction Only slight improvement in global bias, some improvement in regions
July 2001 T2m Station Observation Differences & Improvement
July 2001 T2m Station Observation Amplitude Differences & Improvement Amplitude of the monthly mean diurnal cycle The daily bias correction has little or negative affect on amplitude The diurnal bias correction makes a positive impact for many of the largest biases
July 2001 T2m Station Observation Amplitude Differences & Improvement Most regions are improved, though to varying degrees Eastern Europe is a region where the diurnal amplitude is not improved by diurnal bias correction Daily correction seems to improve the amplitude in Eastern Europe
Fluxes & Soil Moisture
EOP-1 July Sept, 2001 Reference Sites mostly in GEWEX CSEs 40 stations in total 8 with complete energy/radiation observations (Circled) CEOP In-situ Observations
CEOP July 2001 In-situ mean Biases Station Q2m T2m Ts Rn Hs LE Hg Rsd Rsu Rld Rlu B BALTEX Cabauw -1.94 3.10 24.3 59.7-48.0 62.7 7.8-18.8 14.8-47.9 BALTEX Lindenberg -0.56 0.14 45.9 34.9-14.6 0.9 52.6 10.7-5.8-0.9-41.2 MAGS BERMS -1.38 0.31 5.5 8.1-25.6 6.0 39.3 19.5-2.1 12.1-31.4 LBA Manaus -0.91-1.15 2.8-2.3-14.3 3.0 8.4-4.5-4.6-17.4 LBA Rondonia -1.99 0.31-0.74 2.9 5.6 4.5 25.8 9.3-20.4-6.0-33.2 GAPP Bondville -1.05 1.88 1.65 30.1 13.7-17.9-1.2 18.0-1.4 11.1 20.4-30.2 GAPP Ft_Peck -1.75-1.26-4.05 25.8 19.9-4.6 28.9 15.0 13.5-13.2-2.9 GAPP ARMSGP (E15) 1.76-1.32-2.68 30.5 8.9 37.2 2.4 11.4-13.7-10.6-16.6 4.0 ARM NSA Atqasuk -0.73-1.15-0.94 15.8-40.4 ARM NSA Barrow 0.37 4.42 2.07-30.5-40.4 CAMP Mongolia (230) -0.24 0.41-0.31 18.2 8.2 g/kg K W m -2 Mean biases of the diurnal bias correction Redshows improvement of the mean bias (Blue shows degradation)
July 2001 CEOP Comparisons CTL LBA Rondonia Site (-10.08, -61.93º) Observations are dotted lines Improvement in the sensible & latent heat flux due to the DBC EXP2 Significant reduction in the daytime warm bias in the skin temperature and 2 m temperature with inclusion of DBC. Ground heat flux is the residual from the energy balance and thus includes the energy flux from the bias term.
July 2001 CEOP Comparisons CTL GAPP Bondville Site (40.01, -88.29º) Observations are dotted lines Slight improvement shown in the sensible heat flux, but not the latent heat flux. EXP2 With reduced sensible heat there is a compensating increase in the residual ground heat flux, but this is likely attributable to the bias flux which was not removed from the residual. Only small improvement in the TSKIN and T2m resulting from the DBC.
Conclusions Preliminary results show that the coupled skin temperature assimilation and bias correction was for the most part a success The analysis closely draws to the observations, and the effect of the coupling reduces the bias in the background skin temperature by ~ 1 K While diurnal bias correction does not improve the mean temperature much over daily, but the amplitude is generally improved Having coordinated flux stations at many different climate regimes greatly improves the validation of the surface energy budgets
Future Work Long term assimilation desired for CEOP EOP3 and EOP4 The degradation of latent heat flux was ~10Wm-2 at two CEOP stations, currently looking at the relationships between Ts bias and soil moisture bias (nonlinear and less than ideal) There are ongoing experiments in the GMAO to assimilate clouds and also soil moisture, a multi-variate assimilation would be more desirable ISCCP presently does not extend past Sept 2001, MODIS is available for Aqua and Terra, but only 4 times/day